Bank Stress Tests and Other Acts of Faith

rcwhalen's picture

First it needs to be said that the Fed deserves credit for conducting these stress tests, flawed as they may be.  Would that the Fed had followed policies over the past 30 years that made this exercise unnecessary.  But the fact of a standardized annual stress regime is not a bad thing.  The difficulty comes because economists are involved.

The first rule of financial analysis is that you never mix apples and organges, namely financial analysis and economic guesswork.  The task of analytics is to stress the enterprise in an operational sense, then comes the job of trying to relate these observed limits with the outside world.  Attempts to plumb econometric analysis into financial models are alway bound to create more noise than anything else.  But of course the Fed is run by economists and they cannot help themselves when it comes to building models.

With that caveat, let's compare the Fed stress test results with the Economic Capital ratings produced by my partner Dennis Santiago at our ratings firm IRA.  Unlike the econometric mumbo jumbo produced by the Fed as the framework of assumptions for the stress tests, at IRA we use the balance sheet alone to generate a maximum probable loss for three buckets: trading, lending and investing.

The sum of these three stressed MPL estimates is designated Economic Capital or "EC."  Divide income by EC and you have Risk Adjusted Return on Capital or "RAROC."  We then compare EC with Tier 1 Risk Based Capital ("RBC") and produce a ratio between how much capital the bank has vs. the risk-based world of EC.

Then we compare this risk-based output with the public data Bank Stress Index ("BSI") and CAMELS ratings published each quarter by IRA.  The table below show the BSI letter ratings from The IRA Bank Monitor for many of the stress test group banks as of December 31, 2011 and the ratio of EC to T1RBC.  Foreign BHC ratings are for the US units only. 
           
 
IRA Bank Monitor -- Q4 2011



Name Rating BSI Score EC:T1RBC
JPMORGAN CHASE & CO. A 1.38 5.845
BANK OF AMERICA CORPORATION B 1.69 1.519
CITIGROUP INC. B 1.96 2.709
WELLS FARGO & COMPANY A 1.23 1.557
U.S. BANCORP A 1.21 1.17
PNC FINANCIAL SERVICES GROUP,
INC
A 1.2 1.343
BANK OF NEW YORK MELLON
CORPORATION
A+ 0.88 2.99
STATE STREET CORPORATION A+ 0.97 6.094
HSBC HOLDINGS PLC B 1.95 2.913
CAPITAL ONE FINANCIAL
CORPORATION
C 2.12 1.635
TORONTO-DOMINION BANK B 1.74 4.407
BB&T CORPORATION A 1.37 1.289
SUNTRUST BANKS, INC. B 1.85 0.868
UK FINANCIAL INVESTMENTS
LIMITED
C 2.09 0.614
REGIONS FINANCIAL CORPORATION C 2.48 0.629
FIFTH THIRD BANCORP A 1.28 0.763
GOLDMAN SACHS GROUP, INC. A+ 0.63 2.225
NORTHERN TRUST CORPORATION A+ 0.94 1.345
BANK OF MONTREAL B 1.67 0.63
MORGAN STANLEY DELTA HOLDINGS
LLC
A+ 0.82 0.113
AMERICAN EXPRESS COMPANY C 2.31 0.987
Industry Benchmark Year (1995) A 1 NM

    
Source: FDIC (RIS)/The IRA Bank Monitor        

If the bank has an EC to T1RBC ratio much greater than 1, the model is telling you that the risk taken by the enterprise exceeds the ability of the nominal regulatory capital to support it.  Notice that JPMorgan and not Citigroup ("C") is the outlier when you look at the group based upon EC, followed by State Street ("STT"), Toronto Dominion's US units and Goldman Sachs ("GS").   Note that the IRA ratings are a mechanical survey with all the participants treated the same.

So when I look at the Fed stress tests, which seem to be the result of a mountain of subjective inputs and assumptions, the overwhelming conclusion is that these tests are meant to justify past Fed policy.  Policy like alowing large banks to pay dividends and, especially, move loan loss reserves back into income by most of the TBTF banks.  Significantly, US Bancorp and smaller institutions have generally not played this dangerous game with loan loss reserve releases back into income because they could easily cover the dividend.

One of the first things to notice about the conflict between economic and analytics is the mismatch between the HPI series in the Fed stress assumptions and the losses allocated to second lien loans.  The $56 billion in total losses for all participating banks (Figure 9) is way, way too small if HPI is going to fall another 20%.  Try more like $200 billion.  In fact, you could haircut Wells Fargo and C second lien portfolios by $50 billion each today given where the HPI stands.

But as we have written over the past several weeks in The Institutional Risk Analyst, the Fed does not want to believe that there is a problem with real estate.  As my friend Tom Day wrote for PRMIA's DC chapter yesterday: "It remains hard to believe, on the face of it, that many of the more damaged balance sheets could, in fact, withstand another financial tsunami of the magnitude we have recenlty experienced and, to a large extent, continue to grapple with. "

http://riskmanagementcommunity.com/message/1284#1284

Then again, look at the mere $62 billion loss on first lien mortgages in the supervisory stress scenario.  Since real estate is half the total $13 trillion balance sheet of the US banking system and more like 3/4 of total exposure if you include RMBS, how does the Fed manage to keep total real estate losses below $150 billion in the stressed scenario?   Again, the Fed party line is that there is no problem with real estate.  But to be fair to the Fed, they are looking at the same baked and spun disclosure from the TBTF banks as we all.  There were reports that the Fed initially wanted to ignore the 2009 FASB rule changes in fair value of illiquid assets, but banks pushed back and insisted on "new GAAP" as the rule.

The big eye opener which is NOT a surprise in the stress scenario, at least if you've been paying attention, is watching C catching up with Capital One Financial ("COF") in terms of total loan loss rates (Figure 10).  C is clearly the outlier among the top four zombie banks, but to see the loss rate for C catch COF in the stress simulation is really mind boggling.  First, it just confirms our view that the C business model remains decidely subprime.  But also suggest that in a stress scenario, such distinctions start to blur.

The median loss rate for the whole group of 7%.1 percent is bad enough and well above losses reached during the recent crisis, but look at Amex, Bank America ("BAC"), Regions "RF") and Wells Fargo ("WFC") all well above the line.   And then look at WFC as the second leading loss rate institution behind C in first liens.  This confirms my long held view that the aggressive accounting and lending practices at WFC are masking a far more risky franchise than Warren Buffet and others like to believe.  Or as I said on "Fast Money" this week, the REO hidden inside banks like BAC and WFC is probably equal to what they show us on the books today. 

When you get to junior liens and HELOCs (Figure 13) you will understand why I have been saying that Ally Financial and BAC need to be restructured.  With a plus 20% loss rate on second liens, Ally has substantial capital issues to put it mildly.  But look at C right behind them with a loss rate in the mid-teens followed  by BAC.  Yikes.  This type of loss rate is typical for credit cards and both of these second lien portfolios are > $100 billion. 

And the real lesson, dead friends, is that the good old USA is a subprime nation, a society of individuals whose aggregate propability of default is probably around a "B" to "CCC."  Covert the loss rates in the stress tests to bond ratings using the break points from Moody's or S&P and tell me what you see.

Last point on Ally Financial:  Yikes.  Probably weaknes results of whole group.  Memo to POTUS: File Ch 11, sell auto biz, bank to GM in 365 sale.  Liquidate ResCap.  Declare success.  But do not be surprise if BAC follows if Ally goes into bankruptcy.  The one thing that the Fed almost completely ignores is the vast financial risk facing BAC and Ally, and to a lesser degree, WFC, JPM and C.

While on Page 37 the Fed states that "PPNR projections also incorporate the incremental impact of the recent mortgage related settlement between several of the CCAR BHCs and state and federal authorities," but this does not include estimates for put-back and fraud claims that are orders of magnitude larger than the foreclosure settlement.  If there is one key area where the Fed analysis reveals some fundamental weakness is an appreciation for the size and proximyty of some substantial financial and reputational risks to many banks in RMBS litigation.

See you on CNBC "Squawk Box"  ~ 8:30 tomorrow.